TsinghuaAI/CPM-2-Pretrain

Issues

• 数据开源

  #32 opened 2 years ago by joan126
  1

• help 我单机测试两台机器都能正常，但是多机器并行后会出现环境问题

  #23 opened 3 years ago by XiaoqingNLP
  2

• fatal error：pybind11/pybind11.h:no such file or directory

  #31 opened 2 years ago by tianjialai
  0

• should docker pull gyxthu17/cpm-2:1.2 be ran under the root dir of this reository?

  #27 opened 3 years ago by lsy641
  3

• MOE 训练问题？

  #28 opened 3 years ago by XiaoqingNLP
  0

• 环境问题

  #29 opened 3 years ago by 520jefferson
  9

• 使用docker 容器运行程序，这个host_file应该如何设置IP？

  #30 opened 3 years ago by XiaoqingNLP
  0

• CPM-2-Pretrain-moe的词表问题

  #17 opened 3 years ago by jiayuchennlp
  2

• 专家网络参数应该broadcast么

  #14 opened 3 years ago by icodingc
  5

• If training, what is the minimum computer configuration

  #11 opened 3 years ago by zhanghongyong123456
  8

• 关于数据处理问题

  #9 opened 3 years ago by RoyZhanyi
  10

• Lack the definition of TransposedSampler in `sampler.py`

  #24 opened 3 years ago by geekinglcq
  1

• about the parameters of MOE?

  #26 opened 3 years ago by XiaoqingNLP
  0

• Model parallelism of CPM2-MoE

  #15 opened 3 years ago by MichaelXSChen
  2

• 数据预处理tokenize无法处理特殊token

  #25 opened 3 years ago by zetian1025
  0

• CPM2模型推理代码

  #22 opened 3 years ago by Bournet
  2

• How do I start and use the docker image? Is a k80 enough?

  #7 opened 3 years ago by MikeyBeez
  4

• 关于CPM2模型生成的问题

  #16 opened 3 years ago by jiayuchennlp
  9

• Dev/Test split for Math23K dataset

  #3 opened 3 years ago by sjy1203
  3

• Link for 《CUGE: A chinese language understanding and generation evaluation benchmark》

  #2 opened 3 years ago by sjy1203
  3

• 如何加载CPM2.1 或者cmp2.0 模型进行微调？

  #21 opened 3 years ago by XiaoqingNLP
  0

• How to use BMInf to inference 100000.tar 11B model?

  #20 opened 3 years ago by linjianz
  1

• 文本生成强化版模型 CPM2.1的链接是哪个？

  #18 opened 3 years ago by chenjunqiang
  1

• CPM-2-Pretrain是否没有transformer版本？

  #19 opened 3 years ago by NemoCoder
  1

• CPM-2-Pretrain数据处理与读取问题

  #10 opened 3 years ago by RoyZhanyi
  1

• Deepspeed Zero3?

  #12 opened 3 years ago by k15201363625
  5

• Model checkpoint convert

  #13 opened 3 years ago by k15201363625
  2

• The import path is not correct.

  #8 opened 4 years ago by xia-xiao
  1

• 您好，CPM-2技术报告的链接404了

  #6 opened 4 years ago by wakafengfan
  1

• NA

  #5 opened 4 years ago by ShenDezhou
  0

• 请问后续会放出sample供使用参考吗？

  #4 opened 4 years ago by NLPIG
  3

• could you give the link of your cpm-v2 paper?

  #1 opened 4 years ago by world2vec
  1